Nonlinear Modeling And Controller Design Of A Small-scale Unmanned Helicopter

The small-scale unmanned helicopter has wide prospective of applications in many areas, and is being extensively studied all over the world. This dissertation investigates the nonlinear modeling and controller design of small-scale helicopters and develops an experimental setup to validate the model and the controllers. The ultimate goal is to develop an autonomous flight control system that maximizes the performance of small-scale unmanned helicopters as well as guarantees the flight safety. The main contribution of the dissertation is as follows:First, we developed a nonlinear model of a small-scale helicopter near hovering. On the basis of the Blade Element theory and Rotor Disk model, the aerodynamics and the flapping dynamics of the main rotor are modeled. Then, the main rotor inflow dynamics is modeled based on the three-state nonlinear Pitt/Peters model. To characterize the model helicopter, the dynamics of the stabilizer bar and the tail rotor with the AVCS (Angular Vector Control System) electronic gyro are also studied in detail.Next, we developed a nonlinear model of the helicopter on a test bench when the flapping states of the main rotor and the stabilizer bar are considered. The test bench is constructed for experimental testing of the attitude control of the small-scale helicopter. The unknown model parameters are estimated using the EKF (Extended Kalman Filter) with flight test data of the helicopter operating on the test bench. Then, it is proved that the nonlinear model can be globally linearized using the dynamic feedback linearization technique. A heuristic strategy is proposed to reduce the effect of the saturation actuators on the closed-loop performance. Moreover, a robust performance criterion for the experimental system is introduced. Using the Convex Integrated Design (CID) method, it is possible to design a single closed-loop controller that satisfies a set of multiple conflicting performance specifications. However, direct use of the CID method leads to a controller with complicated form which is not suitable for real-time implementation on the helicopter platform. So, we extended the standard CID method to a more general control system framework to solve the conflicting simultaneous performance design problem, which is referred to as ECID (Extended Convex Integrated Design) method. Compared with the standard CID design, the ECID procedure generates a relatively simple controller. Finally, the synthesized controller is tested by simulations and is validated on the experimental small-scale test helicopter. The experimental results demonstrate the expected performance of the proposed controller.Finally, in order to design a nonlinear controller for small-scale autonomous helicopters, an integrated nonlinear model of a small-scale helicopter for hovering control is developed, and the unknown parameters in the nonlinear model are estimated using the system identification method. To estimate the states in the nonlinear system by using the IMU and GPS sensors, an UKF (Unscented Kalman Filter) observer is designed, whose validity is confirmed by the flight experiments. It is demonstrated that the full nonlinear model can be converted into a controllable linear system via the dynamic feedback linearization technique, and its nonlinear attitude subsystem has a stable zero dynamics. In order to avoid the potential danger caused by the singularities in the feedback linearized system, a cascade control structure is proposed. Finally, a robust controller is designed by applying the ECID method to the cascade system, and simulations are carried out to show the good performance of the controller.